From 8cae6f79d1a3df3db8a097f59bbe67fff54e143a Mon Sep 17 00:00:00 2001 From: Aleksei Sviridkin Date: Tue, 21 Apr 2026 00:05:56 +0300 Subject: [PATCH] fix(tests): increase Bootstrap Talos cluster timeout to 60s On fast ephemeral runners a freshly booted Talos node has not yet converged NTP when the previous 10s outer timeout fired, so the embedded 'until talosctl bootstrap; do sleep 1; done' retry loop only got one 5-6s attempt before the loop was killed. Every RPC returned 'FailedPrecondition: time is not in sync yet' and E2E failed in the Prepare environment step. Bump the outer timeout to 60s and the inter-attempt sleep to 2s. This aligns with the surrounding bootstrap gates (etcd members check at 180s, drain RPC errors at 60s, node registration at 60s) and gives NTP plenty of time to converge on ephemeral runners. Assisted-By: Claude Signed-off-by: Aleksei Sviridkin --- hack/e2e-prepare-cluster.bats | 7 +++++-- 1 file changed, 5 insertions(+), 2 deletions(-) diff --git a/hack/e2e-prepare-cluster.bats b/hack/e2e-prepare-cluster.bats index df90b91a..7c5e85ff 100644 --- a/hack/e2e-prepare-cluster.bats +++ b/hack/e2e-prepare-cluster.bats @@ -235,8 +235,11 @@ EOF } @test "Bootstrap Talos cluster" { - # Bootstrap etcd on the first node - timeout 10 sh -ec 'until talosctl bootstrap -n 192.168.123.11 -e 192.168.123.11; do sleep 1; done' + # Bootstrap etcd on the first node. + # Retry for up to 60s: talosctl bootstrap refuses with "time is not in + # sync yet" until NTP converges on a freshly booted node, which on fast + # ephemeral runners can take several attempts of ~5s each. + timeout 60 sh -ec 'until talosctl bootstrap -n 192.168.123.11 -e 192.168.123.11; do sleep 2; done' # Wait until etcd is healthy if ! timeout 180 sh -ec 'until talosctl etcd members -n 192.168.123.11,192.168.123.12,192.168.123.13 -e 192.168.123.10 >/dev/null 2>&1; do sleep 1; done'; then